Discovering sporting talent has historically depended on local scouts, expensive testing infrastructure, and access to formal training. Many capable young athletes remain unseen because they lack proximity to scouting networks or facilities. Advances in artificial intelligence, computer vision, and the near- universal presence of smartphones open the possibility of fair, scalable, and data-driven talent identification. This paper presents the design, implementation, and evaluation of a mobile-first AI platform that measures sport-specific performance using only a smartphone camera and lightweight on-device models. Our system emphasizes human-centered lalnguage, actionable feedback, and accessibility across socio-economic backgrounds. We describe the technical pipeline — from data collection, pose estimation and feature extraction to scoring, personalization and discovery — and demonstrate the platform’s performance in pilot trials across athletics, football, basketball, and cricket. We also discuss ethical considerations, inclusion strategies, and deployment pathways for educational and sports organizations. The goal is to enable young athletes everywhere to access objective, growth-oriented assessment without requiring expensive hardware or gatekeeping.
Gupta et al. (Wed,) studied this question.